The Research On Control System Of TIG Welding Pool Width With Vision Sensing Based On XPC Target

To get the accurate system model and control arithmetic of complex welding process is very difficult,and the results of researching complex welding control process are less reliability by usual method of simulation.So the Real-time(RT) simulation method's significance should be pay attention,studying the control process of target system directly in the hardware loop without any model and arithmetic by used the RT simulation method is feasible.Some key problems were studied in this paper to try to investigate control complex welding process with RT simulation methods.Firstly,the RT simulation experiment of the current closed-loop PID control for the TIG welding process was performed with the RT simulation platform constructed on the basis of Matlab/Simulink of xPC target module,In the experiment,the optimum control parameters were selected with the method of trial and error,and the square and complex waves on the range from 1.0 Hz to 0.1 Hz were simulated for online control. It was recognized from the experimental result that optimum control parameters could be obtained with online simulation on the basis of xPC target module in Matlab/Simulink and the RT simulation method was high speed of compute and perfect real-time capability.Secondly,another vision sensing subsystem for taking and processing the image of the pulsed TIG welding pool has been added in the current closed-loop PID control platform,furthermore,another experiment of the width of welding pool fuzzy control with this platform by used RT simulation based on xPC Target module in MATLAB/SIMULINK was performed.In detail,obverse width of welding pool was inspected in real-time;accordingly the peak value of pulsed current was adjusted with the purpose of getting good welding formation.In this experiment some key problems were solved,one problem was choice of the frame in which width of welding pool could be picked up well and truly,and the solution was that the strict relationship between sum of gray value of welding zone image and current value had been utilized in order to pick-up the first frame in base current period;another problem was the edge of welding pool detection,it can be resolved by just extracted the welding pool zone image besides by utilized those ordinary edge detection algorithms,and it could improve not only the computational efficiency but also the precision to get the width of welding pool through reducing interference from arc in the welding zone image;the third key problem was the design of fuzzy control arithmetic,in this paper a Sugeno model was adopted which include the deviation and rate of change of deviation in real time,as double-input and increment of peak value of current as single-output.In the end two experiments of welding on echelon plate with different condition of heat exchanging were executed in order to inspect the effect of this RT fuzzy control system. And the result as follows:one test plate was used the changeless welding parameters, its width of welding beam gradually increased because the gradually deteriorate condition of heat exchanging until the welding pool appeared collapse in trivial,then the beam width became smaller,the max deviation was 3mm.another test plate was used RT fuzzy control system,the fluctuation range of welding pool width was±1mm.